US5247467A  Multiple variable compensation for transducers  Google Patents
Multiple variable compensation for transducers Download PDFInfo
 Publication number
 US5247467A US5247467A US07/394,617 US39461789A US5247467A US 5247467 A US5247467 A US 5247467A US 39461789 A US39461789 A US 39461789A US 5247467 A US5247467 A US 5247467A
 Authority
 US
 United States
 Prior art keywords
 sv
 value
 input
 reference
 supplemental
 Prior art date
 Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
 Expired  Fee Related
Links
 229910019946 SS Inorganic materials 0 description 2
 229910019939 S—S Inorganic materials 0 description 2
 238000005284 basis set Methods 0 description 2
 238000004422 calculation algorithm Methods 0 description 3
 230000001010 compromised Effects 0 description 1
 230000000875 corresponding Effects 0 claims description 13
 230000000694 effects Effects 0 claims description 6
 230000001747 exhibited Effects 0 description 2
 238000007667 floating Methods 0 description 2
 230000014509 gene expression Effects 0 claims description 3
 238000007689 inspection Methods 0 description 1
 239000007788 liquids Substances 0 description 3
 238000007620 mathematical function Methods 0 description 3
 239000011159 matrix materials Substances 0 description 10
 230000015654 memory Effects 0 description 1
 238000000034 methods Methods 0 description 10
 238000009740 moulding (composite fabrication) Methods 0 claims 10
 229910052698 phosphorus Inorganic materials 0 description 2
 239000000047 products Substances 0 description 1
 229910052705 radium Inorganic materials 0 description 1
 230000004044 response Effects 0 description 1
 230000000717 retained Effects 0 description 1
 239000004065 semiconductor Substances 0 description 1
 230000035945 sensitivity Effects 0 description 10
 230000003595 spectral Effects 0 description 1
 230000003068 static Effects 0 description 1
 238000003860 storage Methods 0 description 3
 229910052717 sulfur Inorganic materials 0 description 1
 230000000153 supplemental Effects 0 abstract claims description 103
 239000011135 tin Substances 0 description 1
 229910052718 tin Inorganic materials 0 description 1
 229910052720 vanadium Inorganic materials 0 description 2
Images
Classifications

 G—PHYSICS
 G01—MEASURING; TESTING
 G01D—MEASURING NOT SPECIALLY ADAPTED FOR A SPECIFIC VARIABLE; ARRANGEMENTS FOR MEASURING TWO OR MORE VARIABLES NOT COVERED IN A SINGLE OTHER SUBCLASS; TARIFF METERING APPARATUS; MEASURING OR TESTING NOT OTHERWISE PROVIDED FOR
 G01D3/00—Indicating or recording apparatus with provision for the special purposes referred to in the subgroups
 G01D3/02—Indicating or recording apparatus with provision for the special purposes referred to in the subgroups with provision for altering or correcting the law of variation
 G01D3/022—Indicating or recording apparatus with provision for the special purposes referred to in the subgroups with provision for altering or correcting the law of variation having an ideal characteristic, map or correction data stored in a digital memory

 G—PHYSICS
 G01—MEASURING; TESTING
 G01D—MEASURING NOT SPECIALLY ADAPTED FOR A SPECIFIC VARIABLE; ARRANGEMENTS FOR MEASURING TWO OR MORE VARIABLES NOT COVERED IN A SINGLE OTHER SUBCLASS; TARIFF METERING APPARATUS; MEASURING OR TESTING NOT OTHERWISE PROVIDED FOR
 G01D3/00—Indicating or recording apparatus with provision for the special purposes referred to in the subgroups
 G01D3/028—Indicating or recording apparatus with provision for the special purposes referred to in the subgroups mitigating undesired influences, e.g. temperature, pressure
 G01D3/036—Indicating or recording apparatus with provision for the special purposes referred to in the subgroups mitigating undesired influences, e.g. temperature, pressure on measuring arrangements themselves
 G01D3/0365—Indicating or recording apparatus with provision for the special purposes referred to in the subgroups mitigating undesired influences, e.g. temperature, pressure on measuring arrangements themselves the undesired influence being measured using a separate sensor, which produces an influence related signal
Abstract
Description
1. Technical Field
This invention relates to compensation for an electrical device that produces a signal representing one output variable, such as a measurement of electrical power, as a function of principal input variable, such as measured voltage and as a function of a plurality of supplemental input variables.
2. Background of the Invention
In the use of an electrical transducer to produce an output electrical variable, such as measured electrical power, from the value of an input electrical variable, such as voltage, the device usually incorporates a sequence of experimentally determined data points relating the input and output variables. These data points, expressed as a graph of input variable versus output variable, define a locus of points that can often be roughly represented by a simple mathematical relationship between the input and output variables. Often, the output variable behaves qualitatively and quantitatively differently for different ranges of values of the input variable. Further, variations in other supplemental variables such as ambient temperature and a frequency that characterizes the input signal also affect the value of the output variable for a fixed input variable value. Variation of the value of the output variable with respect to changes in ambient temperature or in input signal frequency may be expressed by means of a temperature sensitivity parameter and a frequency sensitivity parameter, respectively.
As used herein, a "transducer" will refer to a device that receives an input variable (electrical or nonelectrical), measures the input variable value, and issues a corresponding value of an output variable (electrical or nonelectrical) that characterizes or corresponds to the input variable. As one example, an electrical voltage input variable may be characterized by a corresponding electrical power output variable. As another example, a temperature differential in a device that relies on the Seebeck effect may correspond to an electrical current produced across the Seebeck junction by that temperature differential.
With reference to temperature sensitivity, for many sensors the temperature sensitivity is large and positive for low values of radio frequency power, and is small and is negative for large values of radio frequency power. Further, temperature sensitivity varies significantly from one unit to another of the same model of sensor. This unittounit variation in temperature sensitivity can be as much as a factor of 3 or more for some of the classes of devices tested.
With reference to frequency sensitivity, frequency response is a function of power level as well as a function of ambient temperature. Frequency sensitivity is larger for low values of radio frequency power than it is for high values of radio frequency power. Further, the deviation can be negative for low radio frequency power and positive, but of smaller magnitude, for high radio frequency power.
In the past, others have compensated for the effects of variations in supplemental variables, such as temperature, on device output. In U.S. Pat. No. 4,416,153, issued to Williams, a liquid level monitoring system has a thermistor sensor attached to the tank containing the liquid, and tank temperature is sensed and used to store and produce a linear variation of voltage threshold with tank temperature to produce modified voltage threshold values that indicate liquid level in the tank.
Bozarth et al., in U.S. Pat. No. 4,592,002, disclose a temperaturesensitive digitaltoanalog converter that is provided with a temperature sensing device and the converter output signal is characterized as a function of the input digital data value and temperature. The characterization equation, expressed as a function of the input digital data values with a simple variation with converter temperature, is determined once and stored in a nonvolatile memory.
Temperature compensation for a semiconductor light source is disclosed by Aotsuka et al. in U.S. Pat. No. 4,710,631. Temperature of the light source is sensed, and correction coefficients are computed and stored. These coefficients take into account variations with temperature of radiation intensity and spectral characteristics of the light source. Ambient temperature is sensed and a correction coefficient is applied, without further computation, to maintain approximately constant exposure conditions.
Compensation, in terms of two supplemental variables such as temperature T and static pressure S, of a device that measures a process variable pressure PV as a function primarily of differential pressure P, is disclosed by Malloy in U.S. Pat. No. 4,734,873. The output pressure variable PV is expressed as a polynomial in the differential pressure P, with polynomial coefficients that depend upon the variables S and T. However, little specific information is disclosed on determination of the polynomial coefficients for compensation purposes.
Krechmery et al. disclose a pressure transducer with ambient temperature compensation in U.S. Pat. No. 4,765,188. A temperaturesensitive resistor is deposited on a pressuresensitive diaphragm, and the sensed temperature signal is converted to a digital signal that addresses a preprogrammed data correction table stored in a programmable readonly memory. The numerical data correction is converted to an analog correction signal that corrects the sensed pressure value for ambient temperature of the pressure sensor.
These inventions do not explicitly account for the possibility that the output variable value may behave qualitatively differently for different nonoverlapping ranges of the input variable. These inventions also do not provide for explicit representation of variation of output variable values with changing values of two or more supplemental variables that are independent of the input variable value.
What is needed is a compensation procedure that (1) takes account of the fact that the output variable value may behave differently for different ranges of the input variable value and (2) provides for explicit representation of the variation of output variable value with variation of two or more supplemental variables, such as ambient temperature and input signal frequency, for a fixed value of the input variable.
These needs are met by a compensation method and apparatus that first determines a set of ranges for the input variable, in each of which the value of the output variable may be represented by a different functional relationship to the input variable value. One example is representation of the output variable value by a polynomial in the input variable value, with polynomial coefficients that change from one range to another. Another example is representation by a power law, a trigonometric function, a transcendental function or an orthonormal basis set of functions in different ranges of the input variable.
The inventive approach requires an initial choice of a set of discrete reference temperatures and of a set of discrete reference input signal frequencies or other supplemental variables and determines the functional relationship for a fixed reference temperature and a fixed reference frequency in each of the ranges of the input variable value. For a fixed value of the input variable, this representation is extended by interpolation to all temperatures between the minimum and maximum reference temperatures and to all frequencies between the minimum and maximum reference frequencies. If the functional relationship of output variable and input variable is expressed as a polynomial or other function in each of the ranges of the input variable value, this interpolation may be made by linear or quadratic interpolation of the polynomial coefficients or by interpolation of the resulting value of the output variable.
For a fixed choice of the supplemental variables temperature and frequency, the functional relationships of the output and the input variables may be determined and stored in tables for each range of the input variable value, for future use. The method extends to compensation for each of n(>2) supplemental variables.
A full scale value for the input variable is determined, and a corresponding full scale value for the output variable is determined, using the functional relationships between the output variable and the input variable. The output variable value may then be normalized by division by its full scale value, and the computed output variable value, determined from a prescription of the values of the input variable and the supplemental variables, may be converted from analog form to digital form for subsequent display or further processing.
This approach offers a number of advantages relative to other calibration methods used for electrical or nonelectrical sensors. This approach also offers better accuracy, as compared to hardware interpolation and computation of temperatureinduced deviations. This approach also offers better control of corrections than does a method that provides a single function for representation of the input variable versus output variable functional relationship. If polynomials are used to express this functional relationship, low order polynomials, such as cubic, may be used for representation, which allows for faster computation and allows temperature and frequency interpolation to be done on the polynomial coefficients. The method also allows use of an arbitrary number of ranges of the input variable value, although for some purposes a small number such as four ranges is found to be adequate. The invention provides and algorithm to match coefficients in the functional relationships between adjacent ranges of the input variable value so that the end point values computed in adjacent segments will agree with one another. Finally, the invention extends to compensation for n(>2) supplemental variables.
FIG. 1 is a block diagram of the compensation configuration for two supplemental variables, applied to correction of a calibrated transducer output in accord with the invention.
FIG. 2 is a graphic view, illustrating a possible functional relationship between the values of an input variable V and the values of an output variable P in each of a sequence of ranges of the input variable value.
FIG. 3 is a graphic view, illustrating the relationship between four adjacent pairs or nine adjacent pairs of reference values of two supplemental variables R and S and an arbitrarily chosen temperaturefrequency pair that lies within one of the quadrilaterals defined by the reference value pairs in accord with the invention.
With reference to FIG. 1, the apparatus 11 considered here is a device that receives an input variable v_{in} as an input signal and produces a different output signal in the form of an output variable v_{out}, where the value of the output variable depends upon the value of the input variable and upon values of one, two or more supplemental variables R and S that serve as control input variables for the apparatus 11. Two supplemental variable sensors 21 and 23 sense the present values of the two supplemental variables R and S, respectively. For example, R and S might be apparatus temperature and input signal frequency, as described below. The sensed values of the variables R and S are passed to two range tables 25 and 27 that determine the supplemental variable ranges, such as R_{i} ≦R≦R_{i+1} and S_{j} ≦S≦S_{j+1}, that contain these variables. Each of the variables R and S is determined to lie in a particular one of the ranges for that variable, and the two identified ranges are passed to an interpolation module 29 that contains the coefficients for the reference values (upper and lower limits) for the two ranges that contain the variables R and S. With these reference values in hand, the interpolation module 29 computes four interpolation coefficients, which may be of the forms
R.sub.i+1 R)(S.sub.j+1 S)/(R.sub.i+1 R.sub.i)(S.sub.j+1 S.sub.j), (RR.sub.i)(S.sub.j+1 S)/(R.sub.i+1 R.sub.i)(S.sub.j+1 S.sub.j), (R.sub.i+1 R)(SS.sub.j)/(R.sub.i+1 R.sub.i)(S.sub.j+1 S.sub.j),
and (RR.sub.i)(SS.sub.j)/(R.sub.i+1 R.sub.i)(S.sub.j+1 S.sub.j),
corresponding to the four coordinate pairs (R_{a},S_{b})(a=i or i+1; b=j or j+1). These interpolation coefficients are passed to an input variable transducer 31 to be combined with certain function coefficients a_{k} (R_{a},S_{b}) to produce the interpolated function coefficients a_{k} (R,S) that determine the transducer output variable value v_{out}. This last signal value is then passed to another module 33 (optional) for display or subsequent processing. Alternatively, the function coefficients may be stored in the transducer 31 and passed to the interpolation module 29 whenever needed to form a new group of interpolation coefficients.
One example used here for illustration assigns to the input variable v_{in} the instantaneous or average voltage of an electrical signal and assigns to the output variable v_{out} the electrical power P produced by that input variable. Two supplemental variables mentioned above for this example are ambient temperature T in which the transducer operates and an input signal frequency f that characterizes the time variation of the input variable v_{in}. It should be understood that this choice of input variable, output variable and supplemental variables is made solely for purposes of illustration and that any other reasonable choices of these variables may be made in calibration and compensation of an electrical or nonelectrical device for purposes of subsequent use of the transducer. The compensation method described and claimed here is intended to apply to general functional relationships between an input variable and an output variable, with n(≧2) supplemental variables.
As an illustrative example, consider a transducer that receives an input electrical signal, determines the voltage V of that input signal, and determines and displays the electrical power P associated with that input signal. Assuming that all other relevant physical variables are fixed, FIG. 2 displays a representative functional relationship between the output power P and the input voltage V as determined by the transducer. This functional relationship is generally derived by graphic display and examination of a sequence of data points {(V_{s}, P_{s})}_{s}, as illustrated in FIG. 2, where P_{s} is the observed value of output power corresponding to an observed value V_{s} of the input voltage. The continuous curve shown in FIG. 2 is determined, according to the method of the invention, by choosing a sequence of voltage break points V_{0}, V_{1}, V_{2}, V_{3} and V_{4} so that, within each voltage range V_{r} ≦V<V_{r+1} (r=0,1,2,3), the functional relationship between voltage V and power P is represented by a simple mathematical function such as a polynomial, a power law variation, a transcendental function or other similar function.
The functional relationship within each voltage range is determined by fitting a predetermined mathematical function that has adjustable coefficients or parameters to the subsequence of data points {(V_{s}, P_{s})}_{s} for which the first value V_{s} in each pair (V_{s},P_{s}) lies in the chosen voltage range. The adjustable coefficients or parameters of the mathematical function are chosen so that the portion of the curve shown in FIG. 2 for that voltage range passes through or lies sufficiently close to those data points so that the error in use of the chosen function for representation is less some predetermined normalized error, such as five percent. For the input and output variables V and P, respectively, the approach disclosed here allows determination of a functional relationship in which the normalized error is actually less than one percent. The function coefficients or parameters are readjusted, if necessary, so that the function values at the coincident end points, such as V_{1}, V_{2} and V_{3} of two adjacent voltage ranges are the same for the two functions used in the adjacent voltage ranges. This method is not limited to a choice of four voltage ranges.
Assume, for purposes of illustration here, that the functional relationship within each voltage range shown in FIG. 2 is a polynomial ##EQU1## where the polynomial coefficients ak now define the functional relationship within the chosen voltage range and L is a positive integer. For a transducer relating output power to input voltage, it is often true that the relationship between P and V is linear in the lowest lying voltage range V_{0} ≦V<V_{1} so that the polynomial corresponding to this voltage range has the form P=a_{0} +a_{1} V. However, this feature is not required in the invention.
It is assumed that the mathematical expressions for the functional relationships between the input voltage V and the output power P are determined for each of the voltage ranges shown in FIG. 2, for each of a set of first supplemental variables H_{R} ={R_{1},R_{2}, . . . , R_{M} }(R_{1} <R_{2} <. . . <R_{M}) and for each of a set of second supplemental variables H_{S} ={S_{1},S_{2}, . . . , S_{N} } (S_{1} <S_{2} <. . . <S_{N}), with M and N being fixed integers. For any one of the voltage ranges shown in FIG. 2, the functional relationship between the input voltage V and output power P will vary according to the reference pair (R_{i},S_{j}) (i=1,2, . . . , M; j=1,2, . . . , N) of supplemental variables chosen for the sequence of data points used to determine that relationship. Thus, for each of the M.N reference pairs (R_{i},S_{j}), a set of polynomials shown in equation (1) above, will be determined for the set of voltage ranges shown in FIG. 2.
In order to extend the representation shown in equation (1) to arbitrary values of the supplemental variables R and S that satisfy R_{1} ≦R≦R_{M} and S_{1} ≦S≦S_{N}, the polynomial coefficients ak are interpolated by use of the relationship ##EQU2## whenever a supplemental variable pair (R,S) lies in one of the reference pair quadrilaterals (R_{1} ≦R≦R_{1+1} ; S_{j} ≦S≦S_{j+1}), as illustrated in FIG. 3. The interpolated coefficient a_{k} (R,S) is uniquely determined, although it may be expressed differently, if the coefficient is of the form a_{k} (R,S)=a_{00} +a_{10} R+a_{01} S+a_{11} RS and it is required that a_{k} (R,S)= a_{k} (R_{a},S_{b}) for R=R_{a} and S=S_{b} with a=i or i+1 and b=j or j+1.
Equation (2) for interpolation with respect to two supplemental variables is easily extended to interpolation with respect to a larger number n of supplemental variables. For example, in order to compensate for n=3 supplemental variables, denoted (Q,R,S) with Q_{h} ≦Q≦Q_{h+1}, R_{i} ≦R≦R_{i+1} and S_{j} ≦S≦S_{j+1}, the equation (2) would be replaced by the relationship ##EQU3## Three suitable supplemental variables might be ambient relative humidity, ambient temperature and input signal frequency. The apparatus illustrated in FIG. 1 for n=2 supplemental variables extends to n>3 supplemental variables, in a manner that should be clear to one of ordinary skill in the art.
One of the advantages of using the polynomials shown in equation (1) above to represent the functional relationship between input voltage V and output power P is that the polynomial coefficients themselves are the only quantities that change with variation of the supplemental variables R and S here. One can, of course, replace the power law variation of the individual terms (v_{in})^{k} shown in equation (1) above by a more general prescription ##EQU4## where the multiplier coefficients c_{k} are adjustable and the functions F_{k} (V) are power law functions, trigonometric functions, transcendental functions, an orthonormal basis set of functions or other wellbehaved functions, chosen to provide a good approximation for P as a function of the input variable V in eq. (4). The coefficients c_{k} (Q_{a},R_{b},S_{c}) in equation (4) may be extended to general values of the supplemental variables Q, R and S by replacing the polynomial coefficients a_{k} (Q_{a},R_{b},S_{c}) in equation (3) by the function coefficients c_{k} (Q_{a},R_{b},S_{c}) with a=h or h+1, b=i or i+1 and c=j or j+1. The function F_{k} (v_{in}) might be chosen to be a simple power law function (v_{in})^{k}, a korder orthonormal polynomial such as a Legendre, Laguerre or Hermite polynomial, or a sinusoidal function such as sin[k v_{in} +φ] or cos[k v_{in} +φ] for k= 0,1, . . . , L. However, if the functions F_{k} (V) are allowed to depend on internal parameters that change with variation of the supplemental variables R and S, it will in general not be possible to interpolate the variation of output power P by merely interpolating the multiplier coefficients c_{k} in equation (3). The method of compensation discussed thus far allows representation of the output power P from a sensor in terms of the primary input variable, which is the voltage V of the input signal, and of supplemental variables such as ambient temperature T and characteristic input signal frequency f. These considerations are easily generalized to determination of the representation of a general output variable v_{out} in terms of a primary input variable v_{in} and a plurality of n(>2) supplemental variables S_{1},S_{2}, . . . , S_{n}, as indicated above.
The linear interpolation procedure for the interpolated function coefficients c_{k} (R,S) can be extended to quadratic interpolation, using the format illustrated in FIG. 3. For the rectangle defined by
R.sub.i ≦R≦R.sub.i+2,
S.sub.j ≦S≦S.sub.j+2,
the general form of a function e_{k} (R,S) that is quadratic in R for a fixed value of S and that is quadratic in S for a fixed value of R is
e.sub.k (R,S)=e.sub.00,k +e.sub.10,k R+e.sub.01,k S+e.sub.20,k R.sup.2 +e.sub.11,k RS+e.sub.02,k S.sup.2 e.sub.21,k R.sup.2 S+e.sub.12,k RS.sup.2 +e.sub.22,k R.sup.2 S.sup.2 (5)
The coefficients e_{mn} are as yet unknown constants that are to be chosen to satisfy the constraints
e(R=R.sub.p,S=S.sub.q)=c.sub.k (R.sub.p,S.sub.q) (p=i,i+1,i +2;q=j,j+1,j+2).(6)
This yields nine linear equations for the nine unknown coefficients e_{mn},k and can be written in matrix form as
M(R.sub.i,R.sub.i+1,R.sub.i+2 ; S.sub.j,S.sub.j+1,S.sub.j+2) E.sub.k =C.sub.k, (7)
where M is a 9×9 matrix, E_{k} is a 9×1 column matrix whose entries are the unknowns e_{00},k, e_{01},k, . . . , e_{22},k, and C_{k} is a 9×1 column matrix whose entries are the known coefficients c_{k} (R_{p},S_{q}) (p=i,i+1,i+2;q=j,j+1,j+2). The determinant of the matrix M vanishes only if at least two of the coordinates R_{i}, R_{i+1}, and R_{i+2} are equal or at least two of the coordinates S_{j}, S_{j+1} and S_{j+2} are equal, which does not occur here. Thus the inverse matrix M^{1} exists, and the formal solution of eq. (6) becomes
E.sub.k =M(R.sub.i,R.sub.i+1,R.sub.i+2 ;S.sub.j,S.sub.j+1,S.sub.j+2).sup.1 C.sub.k. (8)
The matrix M will be referred to here as the quadratic interpolation matrix and has the entries 1, R_{p}, S_{q}, R_{p} ^{2}, R_{p} S_{q}, S_{q} ^{2}, R_{p} ^{2} S_{q}, R_{p} S_{q} ^{2} and R_{p} ^{2} S_{q} ^{2} (p=i,i+1,i+2; q=j,j+1,j+2) for any row of the matrix M. The solution E_{k} shown in eq. (8) is unique so that any other method of determining the coefficients e_{mn} in eq. (5) for which the nine equations (6) are satisfied, will be identical to the coefficients e_{mn} determined from eq. (8).
A second method of determining the coefficients e_{mn} in eq.(5) that satisfy the nine equations (6) is to construct a sum of nine terms for the function e(R,S) so that precisely eight of these nine terms vanish for each choice (R,S)=(R_{p},S_{q})(p=i,i+1,i+2;q=j,j+1,j+2) of the pair of coordinates, and the ninth surviving term that does not vanish is equal to a_{k} (R_{p},S_{q}). It can be verified by inspection that the following choice satisfies eq. (6) is quadratic separately in the variables R and S: ##EQU5## where the symbol ##EQU6## is the usual product symbol. This solution of the nine eqs. (6) has the advantages that it explicitly exhibits the solution and no matrix inversion is required.
The coefficients c_{k} (R_{i},S_{j}) may be interpolated linearly with respect to both supplemental variables R and S, may be interpolated linearly with respect to one variable and quadratically with respect to the other variable, or may be interpolated quadratically with respect referred to collectively herein as interpolation "at most quadratically" with respect to the two variables.
Preferably, the input variable break points V_{r} (r=0,1,2,3,4 in FIG. 2) are chosen and fixed for all transducers of a given type or model. The break points may be chosen so that the amount of curvature exhibited by the continuous curve in FIG. 2 that represents the input variableoutput variable relationship, is approximately the same in each range. For the example above, this insures that the value of output power P with input voltage V will not vary too quickly in any one of the voltage ranges. However, other criteria may also be used to determine the break points.
Where a sequence of voltage ranges V_{m} <v_{in} <V_{m+1} (m=0,1, . . . , M1) is used, a problem may arise concerning matching of the function values v_{out} at a common end point or break point of two adjacent voltage ranges. For example, if the polynomial representations
v.sub.out.sup.() =a.sub.0 +a.sub.1 v.sub.in +a.sub.2 (v.sub.in).sup.2 +a.sub.3 (v.sub.in).sup.3 (V .sub.m1 ≦v.sub.in ≦v.sub.m),(12)
v.sub.out.sup.(+) =b.sub.0 +b.sub.1 v.sub.in +b.sub.2 (v.sub.in.sup.2)+b.sub.3 (v.sub.in).sup.3 (V .sub.m ≦v.sub.in ≦V.sub.m+1) (13)
are used, it may occur that these two representations do not agree at the common end point v_{in} =V_{m} so that
v.sub.out.sup.(+) (v.sub.in =V.sub.m)v.sub.out.sup.() (v.sub.in V.sub.m)=Δ.sub.m ≠0. (14)
The polynomial coefficients are adjusted by replacing the coefficients a_{x} and b_{y} by new coefficients that are defined by
a.sub.0 '=a.sub.0,
a.sub.1 '=a.sub.1,
a.sub.2 '=a.sub.2 Δ.sub.m /4V.sub.m.sup.2,
a.sub.3 '=a.sub.3 +Δ.sub.m /4V.sub.m.sup.3,
b.sub.0 '=b.sub.0 Δ.sub.m /4,
b.sub.1 '=b.sub.1 Δ.sub.m /4V.sub.m,
b.sub.2 '=b.sub.2,
b.sub.3 '=b.sub.3. (15)
If this approach is followed for each of the common voltage range endpoints, the resulting differences v_{out}.sup.(+) (v_{in} =V_{m})v_{out}.sup.() (v_{in} =V_{m}) are often found to be small enough to be ignored. If these adjustments are made serially at each voltage range point ("common endpoint") that is shared by two adjacent voltage ranges, it may occur that the coefficient adjustments made with respect to one endpoint are compromised by the coefficient adjustments made at the next consecutive endpoint. In such event, the coefficient adjustment procedure should be repeated, beginning at a common endpoint at one end and moving through consecutive common endpoints toward the opposite end. After a few iterations of this procedure, the differences or errors Δ_{m} at all common endpoints should decrease to acceptably small amounts and the iteration procedure can be terminated. The coefficient adjustment procedure illustrated here is merely one of many such procedures that can be applied.
For the more general case of polynomial representations ##EQU7## with m=1,2, . . . ,p1, define the endpoint error _{m} by the relation
Δ.sub.m =v.sub.out.sup.(+) (v.sub.in =V.sub.m)v.sub.out.sup.() (v.sub.in =V.sub.m). (18)
For any choice of a nonnegative integer M satisfying 0≦M<L, the adjusted coefficients can be determined by the relations ##EQU8## The coefficients a_{k1}.sup.(m)' and a_{k2}.sup.(m)' may be iteratively adjusted for the sequence of common endpoints v_{in} =V_{m} (m=1,2, . . . , p1) until the redefined output values approximately coincide at the common endpoints v_{in} =V_{m}.
In a preferred embodiment in which the supplemental variables are temperature T and frequency f, the number M of reference temperatures is five and the reference temperatures themselves are T=0, 15, 25, 35 and 55° C. In this preferred embodiment the number N of reference frequencies is seven and the reference frequencies are 1, 6, 12, 18, 25 and 26.5 gigacycles/sec. For a representative peak power transducer, four voltage ranges are often sufficient, and the use of cubic polynomials (L=3) in equation (1) above allows a reasonably accurate representation of the functional relationship between input voltage V and output power P, with an associated normalized error of less than five percent in each voltage range defined by two consecutive voltage break points. For a fixed voltage range, chosen temperature and chosen input frequency (not necessarily a reference temperature or a reference frequency), the number of function coefficients needed for variation of input voltage V in that voltage range is 7×5×4×4=560 coefficients. This number of coefficients is stored in a coefficient storage unit such as a PROM or an EPROM or an EEPROM for subsequent transfer to a fast acting random access memory for use by the transducer. If the voltage break points are the same for each device of a given model or type, the voltage break points need only be stored once, also in the coefficient storage unit.
Upon powerup of the transducer, the coefficients and voltage break points are read from the coefficient storage unit and are loaded into a RAM. For a measured ambient temperature T and measured input signal frequency f, a desired full scale output variable value P_{FS} is prescribed and a corresponding input variable value V=V_{FS} is determined for which P(V=V_{FS})=P_{FS}, using eq. (4). The input variable range, V_{o} ≦V.sub. ≦V_{FS} is then divided into 2^{q} equal length intervals, where q is a positive integer, and the definition ##EQU9## is adopted. Each input variable value V in an interval such as V_{m} ≦V<V_{m+1} is assigned a rounded value V_{m}. Power levels P_{m} are computed according to the prescription
P(V=V.sub.m)=P.sub.m=P.sub.FS P.sub.m ', (24)
where P'_{m} is a normalized output variable value satisfying 0≦P_{m} '≦1. Although the levels V_{m} of the input variable are equidistantly spaced, the output variable levels P_{m} or P_{m} ' need not be equidistantly spaced.
The normalized output variable P_{m} ' is now expressed in floating point notation, using r binary digits or bits, so that P_{m} ' is rounded up or down to an rbit representation P_{n} " (n=0,1, . . . , 2^{r} 1) by the prescription ##STR1## The process of rounding up or rounding down will be referred to simply as "rounding".
This rounding procedure is equivalent to the prescription ##STR2## for rounding the nonnormalized output variable value P_{m} for floating point representation. The normalized and rounded output variable values P_{m} " are entered into a transducer lookup table together with the corresponding quantized input variable values V_{m}. Given an input variable value V, the following procedure is followed.
(1) Round V to V_{m} by quantization;
(2) Determine P_{m} =P_{FS} P_{m} '=P(V=V_{m});
(3) Round P_{m} ' to P_{n} " (or P_{m} =P_{FS} P_{m} ' to P_{FS} P_{n} ")
To each input variable value V, the transducer lookup table assigns a rounded value V_{m} and a corresponding rounded output variable value P_{n} " or P_{FS} P_{n} ", and this last value is used for presentation, such as display or subsequent processing. The number r used for rounding of the output variable values P_{n} " is equal to 16 in a preferred embodiment, which produces 2^{16} =65,536 table entries for the output variable values. Another attractive choice is r=8, which produces 256 table entries for the output variable values.
The transducer table relating input voltage V to output power P or to normalized output power P' will be changed whenever the ambient temperature T changes or the input frequency f changes, or both change. A characteristic temperature sensitivity and a characteristic frequency sensitivity may also be determined during interpolation of the coefficients by the relationship in equation (2) above. Minor temperature drift or frequency variation can be accounted for by simply scaling the table entries rather than by complete recomputation of the table, wherever possible. A cache or supplemental memory can be provided to hold entries from the last few sets of table entries computed, and the old table entries can be saved based on use of the Most Recently Used (MRU) algorithm. Under this algorithm, the last few sets of table entries used are retained for possible future use so that the tables are not discarded and recomputed needlessly.
If 16bit accuracy is chosen for subsequent display or processing of a normalized output variable, the dynamic range of such output variable might assign one bit to the sign of the variable and fifteen fractional bits to the magnitude of the value itself. This would allow the normalized output power p to be represented in increments of 0.000122 over a range from 1.0 up to 0.999878. If the output variable will always have one sign (i.e. always be nonnegative or always be nonpositive), the resolution of the output display may be doubled by use of the sign bit as an additional quantization bit.
Claims (18)
v.sub.out =Σα.sub.k.sup.(m) (sv.sub.1.sup.1, . . . , sv.sub.N.sup.1) F.sub.k,m (v.sub.in)
F.sub.k,m (v.sub.in)={v.sub.in)k}(v.sub.in).sup.k1
v.sub.out =ΣΔ.sub.k.sup.(m) (sv.sub.1.sup.1, . . . , sv.sub.N.sup.1) F.sub.k,m (v.sub.in)
F.sub.k,m (v.sub.in)={v.sub.in)k}(v.sub.in).sup.k1
Priority Applications (1)
Application Number  Priority Date  Filing Date  Title 

US07/394,617 US5247467A (en)  19890816  19890816  Multiple variable compensation for transducers 
Applications Claiming Priority (1)
Application Number  Priority Date  Filing Date  Title 

US07/394,617 US5247467A (en)  19890816  19890816  Multiple variable compensation for transducers 
Publications (1)
Publication Number  Publication Date 

US5247467A true US5247467A (en)  19930921 
Family
ID=23559721
Family Applications (1)
Application Number  Title  Priority Date  Filing Date 

US07/394,617 Expired  Fee Related US5247467A (en)  19890816  19890816  Multiple variable compensation for transducers 
Country Status (1)
Country  Link 

US (1)  US5247467A (en) 
Cited By (11)
Publication number  Priority date  Publication date  Assignee  Title 

US5297064A (en) *  19910401  19940322  General Motors Corporation  Sensor lag compensation 
US5369603A (en) *  19920719  19941129  Myers; Allen  Calibration of a nonlinear sensor 
US5373346A (en) *  19910613  19941213  Onset Computer Corp.  Data gathering computer and analysis display computer interface system and methodology 
FR2714174A1 (en) *  19931216  19950623  Bosch Gmbh Robert  Device for operating a signal 
US5440237A (en) *  19930601  19950808  Incontrol Solutions, Inc.  Electronic force sensing with sensor normalization 
US5706214A (en) *  19950329  19980106  Eaton Corporation  Calibration of microcomputerbased metering apparatus 
WO2001050090A1 (en) *  20000105  20010712  Continental Teves Ag & Co. Ohg  Method for establishing a table of correction values and sensor signal and a sensor module 
US6304827B1 (en) *  19990916  20011016  Sensonor Asa  Sensor calibration 
US6321171B1 (en)  19980403  20011120  Tektronix, Inc.  Electronic measurement instrument probe accessory offset, gain, and linearity correction method 
EP1956343A2 (en) *  20070212  20080813  Vaisala Oyj  Method and device for compensating temperature dependencies 
JP2015212664A (en) *  20140506  20151126  株式会社デンソー  Method of manufacturing pressure sensor 
Citations (13)
Publication number  Priority date  Publication date  Assignee  Title 

US4149254A (en) *  19750625  19790410  American Chain & Cable Co., Inc.  Method and apparatus for flow metering 
US4192005A (en) *  19771121  19800304  Kulite Semiconductor Products, Inc.  Compensated pressure transducer employing digital processing techniques 
US4390956A (en) *  19810306  19830628  The Singer Company  Apparatus for correcting measured gas flow 
US4416153A (en) *  19820315  19831122  Innovative Medical Systems, Corp.  Method of compensating a thermistor 
US4437164A (en) *  19810305  19840313  Bristol Babcock Inc.  Ridge circuit compensation for environmental effects 
US4581946A (en) *  19820621  19860415  Masahiro Kanayama  Instrumental error compensation circuit for flow meter 
US4592002A (en) *  19831213  19860527  Honeywell Inc.  Method of digital temperature compensation and a digital data handling system utilizing the same 
US4622843A (en) *  19851227  19861118  Hormel Ronald F  Simplified calibration technique and auto ranging circuit for an electronic compass control circuit 
US4675835A (en) *  19831128  19870623  Peter Pfleiderer  Device for compensating reproduction errors in an electroacoustic transducer 
US4710631A (en) *  19840828  19871201  Fuji Photo Film Co., Ltd.  Temperature compensation for a semiconductor light source used for exposure of light sensitive material 
US4734873A (en) *  19840202  19880329  Honeywell Inc.  Method of digital process variable transmitter calibration and a process variable transmitter system utilizing the same 
US4765188A (en) *  19861124  19880823  Bourns Instruments, Inc.  Pressure transducer with integral digital temperature compensation 
US4873655A (en) *  19870821  19891010  Board Of Regents, The University Of Texas System  Sensor conditioning method and apparatus 

1989
 19890816 US US07/394,617 patent/US5247467A/en not_active Expired  Fee Related
Patent Citations (13)
Publication number  Priority date  Publication date  Assignee  Title 

US4149254A (en) *  19750625  19790410  American Chain & Cable Co., Inc.  Method and apparatus for flow metering 
US4192005A (en) *  19771121  19800304  Kulite Semiconductor Products, Inc.  Compensated pressure transducer employing digital processing techniques 
US4437164A (en) *  19810305  19840313  Bristol Babcock Inc.  Ridge circuit compensation for environmental effects 
US4390956A (en) *  19810306  19830628  The Singer Company  Apparatus for correcting measured gas flow 
US4416153A (en) *  19820315  19831122  Innovative Medical Systems, Corp.  Method of compensating a thermistor 
US4581946A (en) *  19820621  19860415  Masahiro Kanayama  Instrumental error compensation circuit for flow meter 
US4675835A (en) *  19831128  19870623  Peter Pfleiderer  Device for compensating reproduction errors in an electroacoustic transducer 
US4592002A (en) *  19831213  19860527  Honeywell Inc.  Method of digital temperature compensation and a digital data handling system utilizing the same 
US4734873A (en) *  19840202  19880329  Honeywell Inc.  Method of digital process variable transmitter calibration and a process variable transmitter system utilizing the same 
US4710631A (en) *  19840828  19871201  Fuji Photo Film Co., Ltd.  Temperature compensation for a semiconductor light source used for exposure of light sensitive material 
US4622843A (en) *  19851227  19861118  Hormel Ronald F  Simplified calibration technique and auto ranging circuit for an electronic compass control circuit 
US4765188A (en) *  19861124  19880823  Bourns Instruments, Inc.  Pressure transducer with integral digital temperature compensation 
US4873655A (en) *  19870821  19891010  Board Of Regents, The University Of Texas System  Sensor conditioning method and apparatus 
Cited By (12)
Publication number  Priority date  Publication date  Assignee  Title 

US5297064A (en) *  19910401  19940322  General Motors Corporation  Sensor lag compensation 
US5373346A (en) *  19910613  19941213  Onset Computer Corp.  Data gathering computer and analysis display computer interface system and methodology 
US5369603A (en) *  19920719  19941129  Myers; Allen  Calibration of a nonlinear sensor 
US5440237A (en) *  19930601  19950808  Incontrol Solutions, Inc.  Electronic force sensing with sensor normalization 
FR2714174A1 (en) *  19931216  19950623  Bosch Gmbh Robert  Device for operating a signal 
US5706214A (en) *  19950329  19980106  Eaton Corporation  Calibration of microcomputerbased metering apparatus 
US6321171B1 (en)  19980403  20011120  Tektronix, Inc.  Electronic measurement instrument probe accessory offset, gain, and linearity correction method 
US6304827B1 (en) *  19990916  20011016  Sensonor Asa  Sensor calibration 
WO2001050090A1 (en) *  20000105  20010712  Continental Teves Ag & Co. Ohg  Method for establishing a table of correction values and sensor signal and a sensor module 
EP1956343A2 (en) *  20070212  20080813  Vaisala Oyj  Method and device for compensating temperature dependencies 
EP1956343A3 (en) *  20070212  20130911  Vaisala Oyj  Method and device for compensating temperature dependencies 
JP2015212664A (en) *  20140506  20151126  株式会社デンソー  Method of manufacturing pressure sensor 
Similar Documents
Publication  Publication Date  Title 

Park  Special shielded resistor for highvoltage dc measurements  
CA1242791A (en)  Circuit breaker static trip unit having automatic circuit trimming  
US3535637A (en)  Calibration of electrical measuring transducer devices  
US5329818A (en)  Correction of a pressure indication in a pressure transducer due to variations of an environmental condition  
US4075662A (en)  Method and system for compensating the nonlinearities in a reproduction process  
US4837504A (en)  Electricity meter and method of calibrating same  
EP0110898B1 (en)  Transducer calibration system  
US5440305A (en)  Method and apparatus for calibration of a monolithic voltage reference  
US4765188A (en)  Pressure transducer with integral digital temperature compensation  
KR20100075740A (en)  Mass flow meter and mass flow controller  
US4715003A (en)  Method for temperature compensation and measuring circuit therefor  
US4480312A (en)  Temperature sensor/controller system  
US20050285954A1 (en)  Apparatus and method for determining temperature of an active pixel imager and correcting temperature induced variations in an imager  
US5146788A (en)  Apparatus and method for a temperature compensation of a catheter tip pressure transducer  
US4873655A (en)  Sensor conditioning method and apparatus  
US4536851A (en)  Electronic thermometer and associated apparatus  
US4327416A (en)  Temperature compensation system for Hall effect element  
US5116136A (en)  Temperature measurements using thermistor elements  
US3891391A (en)  Fluid flow measuring system using improved temperature compensation apparatus and method  
CN100351617C (en)  Temperature sensing circuit, semiconductor integrated circuit and regulating method thereof  
US5710370A (en)  Method for calibrating a differential pressure fluid flow measuring system  
US4455095A (en)  System for measuring the heat energy emission of room heating elements  
CN1243954C (en)  Apparatus and method for correcting sensor drift  
US4956795A (en)  Signal conditioners  
US2746295A (en)  Apparatus for measuring and recording pressures indicated by manometer tubes 
Legal Events
Date  Code  Title  Description 

AS  Assignment 
Owner name: HEWLETTPACKARD COMPANY, PALO ALTO, CA A CORP OF C Free format text: ASSIGNMENT OF ASSIGNORS INTEREST.;ASSIGNORS:NGUYEN, BICH;SCHIPPER, JOHN F.;REEL/FRAME:005622/0230;SIGNING DATES FROM 19901106 TO 19901109 

FPAY  Fee payment 
Year of fee payment: 4 

AS  Assignment 
Owner name: HEWLETTPACKARD COMPANY, A DELAWARE CORPORATION, C Free format text: MERGER;ASSIGNOR:HEWLETTPACKARD COMPANY, A CALIFORNIA CORPORATION;REEL/FRAME:010841/0649 Effective date: 19980520 

AS  Assignment 
Owner name: AGILENT TECHNOLOGIES INC., CALIFORNIA Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:HEWLETTPACKARD COMPANY, A DELAWARE CORPORATION;REEL/FRAME:010901/0336 Effective date: 20000520 

REMI  Maintenance fee reminder mailed  
LAPS  Lapse for failure to pay maintenance fees  
FP  Expired due to failure to pay maintenance fee 
Effective date: 20010921 

STCH  Information on status: patent discontinuation 
Free format text: PATENT EXPIRED DUE TO NONPAYMENT OF MAINTENANCE FEES UNDER 37 CFR 1.362 